The Dynamic Effects of Tax Audits

Understanding causes of and solutions to non-compliance is important for a tax authority. In this paper we study how and why audits affect reported tax in the years after audit – the dynamic effect – for individual income taxpayers. We exploit data from a random audit program covering more than 53,000 income tax self assessment returns in the UK, combined with data on the population of tax filers between 1999 and 2012. We first document that there is substantial non-compliance in this population. One in three filers underreports the tax owed. Third party information on an income source does not predict whether a taxpayer is non-compliant on that income source, though it does predict the extent of underreporting. Using the random nature of the audits, we provide evidence of dynamic effects. Audits raise reported tax liabilities for at least five years after audit, implying an additional yield 1.5 times the direct revenue raised from the audit. The magnitude of the impact falls over time, and this decline is faster for less autocorrelated income sources. Taking an event study approach, we further show that the change in reporting behaviour comes only from those found to have made errors in their tax report. Finally, using an extension of the Allingham-Sandmo (1972) model, we show that these results are best explained by audits providing the tax authority with information, which then constrains taxpayers’ ability to misreport.